Data analysis: Picture this

Visualization — the technique for turning reams of complex data into meaningful graphics — is a technology that scientists and engineers have been using for years. But with the computer and Internet revolutions producing an explosion of data for every conceivable issue, visualization should be a technology available to the rest of us.

It can be a powerful aid to understanding. If you've got 100,000 documents to analyze — a mind-numbing task for most people — visualization can be invaluable in showing the relationships between the data in those documents.

A very simple application of information visualization would be a graph of the performance of a company's stock in relation to its earnings over a certain period of time. An application used to represent more complex datasets might use devices such as color, size, shape, proximity and motion to show in two or three dimensions how different datasets relate to one another.

An important element of visualization is the ability of the analyst to "interact" with the data on the screen to see how the relationships change when certain criteria are applied. But for those who don't know anything about the technology, it can be hard to overcome that first experience of visualization.

"Most people do view visualization as just a pretty picture," said Jim Thomas, chief scientist for information technologies at Pacific Northwest National Laboratory. "But what it really provides is a highly interactive framework for information analytics. That's the phrase I prefer to use."

And it's the people part of visualization that might prove more difficult. Unlike scientific visualization, Thomas said, which has "the lovely properties of the physics underneath it," information visualization and data mining — two of the likely mainstream applications — deal with far fuzzier relationships, as well as with the ambiguity of language and words.

"There are major changes in research directions happening as we speak," he said. "The talk over the last 20 years has been about the human/computer interface. Now we speak about the human/ information interface. The issue now is how to engage in a dialogue with information resources, which includes the use of intelligent agents and so on."

It's not enough to focus on how you visualize information, you also have to know how to best represent the information, according to Nahum Gershon, senior principal scientist at Mitre Corp. And not all information is best presented through visualization.

"If you want to tell something about a family and who makes up that family, that can be told in a short and exact way in text," he said. "Trying to tell that in images, you would probably use a family portrait, but why would you assume in the first place that this was a family? We use our own background and conventions, but there is no way from looking at the portrait of knowing that."

In trying to introduce more visualization into the work of government analysts, many of whom are more comfortable with text, it's not as simple as inserting a new technology.

Agencies that adopt visualization will need to devote time to the training of their analysts so that they not only become proficient in the new technology, but also develop trust that the tools will support finding the right solution to an intelligence question, said Jon Dale, research and development program manager in the InnoVision Directorate of the National Imagery and Mapping Agency. By its very nature, visualization involves people.

Despite the potential of visualization, Dale does not believe there will be an "overnight replacement" of current analysis methods.

"Our hope and desire is that visual analysis will become more mainstream, though I'm not sure that will happen," Thomas said. "This involves the human mind and how we explore. The technology is not there yet, but there is enough to at least show what it's capable of."

Robinson is a freelance journalist based in Portland, Ore. He can be reached at hullite@mindspring.com.